This invention relates to line-scan imaging in 3-D, and in particular to the use of an imaging detector such as a single CCTV camera.
Line-scan cameras have been in existence for many years. They consist of a single line of photo sites, usually made of silicon, in front of which is a standard lens controlling focus, aperture etc. This is effectively a one dimensional sensor which can form the image of a scene if relative lateral movement takes place between the object of interest and the sensor. The line-by-line sequence of imaging information is built up into an area image by using a frame store. Typical applications are in checking components being moved by conveyor in a manufacturing environment. These components are usually flat, such as printed circuit boards, where 2-D pattern matching algorithms are used for inspection purposes.
A 3-D (stereoscopic) line-scan system could be built using two line-scan cameras configured to provide left and right perspective information. With such a system it would be possible to extract 3-D coordinate data from the image of an object and thus make 3-D inspection possible. For such a system to work there would need to be precise geometrical alignment of both line-scan cameras and also well matched lenses would be essential. These requirements could cause severe problems in setting up such a system since an image could only be formed after movement has taken place and so real-time adjustments of such simple things as focus are not possible. Since precise matching between the parameters of both cameras is essential if acceptable 3-D images are to be produced, then the set-up procedure can be both time consuming and imprecise.
It is, therefore, an object of the invention in one aspect to avoid these problems.
Therefore, according to the invention apparatus is provided for producing a stereoscopic line-scan image of an object in which a detector camera, such as a television camera, is used to view the object, the detector camera having an output signal generated by scanning the received image in a series of parallel lines across the detector, means are provided to move relative to one another the surface of the object past the area viewed by the detector in a direction which is at right angles to the parallel lines across of the detector, and means are provided to select the outputs from two spaced lines in each picture frame produced by scanning the detector and store those lines in a frame store to build up individual pictures from each of the two lines, so that one individual picture constitutes a left hand picture of a stereoscopic pair and the other individual picture constitutes a right hand image of a stereoscopic pair.
Also according to the invention there is provided a method of producing a stereoscopic pair of images, in which the surface of an object is moved past a detector camera which produces an output signal generated by scanning the received image in a series of parallel lines across the detector, the direction of movement of the object being at right angles to the scanning line direction of the detector, and the outputs from two spaced lines in each picture frame produced by the detector are stored and individual pictures are built up from each of the two lines, so that one individual picture constitutes a left hand picture of the stereoscopic pair and the other individual picture constitutes a right hand image of the stereoscopic pair.
By operating in this way all of the problems of the set up of two separate cameras or detectors are avoided. Thus one only selects the image information from two of the multiplicity of video lines available each frame. If the detector is a camera it can be used as a standard television camera to view the object during set up and its direction, focus, brightness and contrast can be selected to give a good, high quality, image. Then, one can switch to the arrangement according to the invention where the video information from single lines is selected and, as the surface of the object is moved past the television camera, successive signals from those individual lines are selected, stored and then put together to form the two separate pairs of stereoscopic images. All of this can be achieved by means of a suitable program and stored in a frame store and thereafter the image can be looked at and manipulated as required.
By way of example, the left and right hand images which result can be viewed as a stereoscopic pair and there are very well known systems available such as the red/green anaglyph, field sequential, and lenticular screen systems, which enable one to present or view left and right hand images. Therefore any of these systems can be used.
In a more sophisticated operation as will be described below, however, one can subtract the video signal in the left hand image from that in the right hand image, or vice versa. This will have the effect of providing an output signal which is unique. Therefore, one can readily use and store that signal in a data base and compare it with signals obtained from other objects to see if there is correspondence. For example such a technique can be extremely useful in forensic analysis in the comparison of cylindrical objects like spent cartridge cases or in the examination of the bullets after firing as will be described in more detail below.
In order to enhance or reduce the depth of the resulting stereoscopic image when viewed one can, of course, chose lines having varying positions in the overall video picture frame. By choosing lines which are spaced apart as far as possible across the video frame, i.e. a line from the very beginning of each complete scan and a line at the end of such a scan, one can create the greatest impression of depth. By choosing lines which are closer together a reduced impression of depth can be given. This can be achieved by programme means which processes the video signal so that an operator can chose a particular impression of depth which he find suits his purposes the best. The system of the invention is, therefore, very flexible and can present information to the unique requirements of any particular operator yet in a very simple fashion.
Where the detector camera is a television camera, it is being used according to this aspect of the invention, for two distinct functions and such a use is quite unique. Thus the video signal from the television camera can be used as a standard video signal and displayed in standard fashion on a monitor. Alternatively the signal can be processed according to the invention to select individual lines in a scan and the results used to build up a pair of stereoscopic images which can then be viewed or presented stereoscopically.
In another embodiment of the invention X-ray images can be obtained using a point source X-ray and as the imaging chip a full-field digital X-ray plate detector. Such detectors can give an output analogous to the raster scan output from a conventional television camera, the only difference between this arrangement and the use of visible light being that a shadow picture is provided by the X-ray source rather than a light image. The output, however, is a digital signal and individual lines of the signal can be selected to build up the individual picture.
The direction of scanning of the detector must be at right angles to the direction at which the surface of the object moves past it. In normal circumstances with a conventional television camera, this will tend to mean that the scanning lines need to be vertically oriented so that it can look at the surface of an object moving past it horizontally. Traditionally a television camera has horizontal scanning lines and so in one very simple embodiment according to the invention one simply turns the camera through 90xc2x0 so that it""s scanning lines are then substantially upright. Alternatively, however, it is possible to produce by program means upright or vertical lines from the video signal, in which case it would then not be necessary to reorient the camera.
In a particularly favourable embodiment according to the invention the object being examined is a cylindrical object. Under these circumstances the surface of the cylindrical object can be moved past the detector by mounting the object on a horizontally rotatable table and rotating the object about an upright axis with the detector such as a television camera looking at the object and preferably oriented at right angles to its normal orientation so that it""s scanning lines are substantially upright.
The television camera can be in any form but a particularly preferred form is a solid state CCTV camera.
There are several advantages of such a system. One is that since a single lens or in the case of X-rays a single point source, and area array detectors are used, then real-time set-up of an image is very straight forward. Also, the geometric relationship between each line of pixels on the area array device is fixed in the silicon and so any pair of lines of pixels will automatically be oriented in a parallel geometry. By appropriate choice of respective lines, stereoscopic images can then be produced when using the area array in this line-scan mode.
Similar advantages are achieved by using an X-ray point source and a solid state digital imaging plate.
As noted above an important application exists where rotational motion would have significant advantages. The application would be for the inspection of essentially cylindrical objects. Viewing a cylindrical object with a conventional area array camera provides an image which is distorted at its edges due to the affect of linear perspective. This also causes a reduction in resolution at the extreme edges of the cylinder. All-round viewing of such objects is difficult and there are severe problems due to variable reflection across the surface.
Therefore, according to another aspect of the invention there is provided a method of reproducing a the surface of a cylindrical object into a two dimensional image, in which a cylindrical object is rotated about its axis in front of a detector camera, the axis and the orientation of the scanning lines of the detector camera being parallel with one another, and successive line-scans from the detector camera are stored and built up to produce the two dimensional image. Such an arrangement effectively unfurls the surface of a cylindrical object into an area image.
As will be appreciated a system according to the invention as described above using a television camera can be used in this method. To produce a simple two dimensional area image, only a single line from each of the scan images needs to be chosen and the video output from that line is stored in a suitable frame store to create the two dimensional image. Alternatively, and according to another embodiment of the invention, one can reproduce three dimensional information by producing a stereoscopic pair of images. Thus in this case one selects images from two spaced lines in each image scan and then one creates a stereoscopic pair of images as explained above. Thus, for example, the method of the invention can be used to inspect and control production of cylindrical objects.
There is a whole range of essentially cylindrical artifacts which could be inspected in this way. However, one particular problem exists which involves the matching of markings on the sides of spent cartridge cases for forensic ballistics applications. The system according to the invention would have a major advantage for such a situation. Thus a standard mechanism for producing a quality image of the cylindrical surface would then exist.
In particular the output from the detector camera is a line-scan signal which can readily be stored in a frame store and transferred to a data base in a standard manner. The data base can then be built up and searched automatically to match different specimens in much the same way as a fingerprint data base is operated.
A particular advantage of producing a stereoscopic pair of images in this way is that they can be processed to subtract the left image from the right image to provide a unique data signal. This as an alternative can be stored in a data base as noted above and that signal will only be achieved identically in the case where, for example, another spent cartridge casing is fired from the same rifle.
The invention would also have further application in the forensic science of examining the surface markings on fired bullets.